Giorgios Kollias

Network Similarity Decomposition (NSD): A Fast and Scalable Approach to Network Alignment

As graph-structured data sets become commonplace, there is increasing need for efficient ways of analyzing such data sets. These analyses include conservation, alignment, differentiation, and discrimination, among others. When defined on general graphs, these problems are considerably harder than their well-studied counterparts on sets and sequences. In this paper, we study the problem of global alignment of large sparse graphs. Specifically, we investigate efficient methods for computing approximations to the state-of-the-art IsoRank solution for finding pairwise topological similarity between nodes in two networks (or within the same network). Pairs of nodes with high similarity can be used to seed global alignments. We present a novel approach to this computationally expensive problem based on uncoupling and decomposing ranking calculations associated with the computation of similarity scores. Uncoupling refers to independent preprocessing of each input graph. Decomposition implies that pairwise similarity scores can be explicitly broken down into contributions from different link patterns traced back to a low-rank approximation of the initial conditions for the computation. These two concepts result in significant improvements, in terms of computational cost, interpretability of similarity scores, and nature of supported queries. We show over two orders of magnitude improvement in performance over IsoRank/Random Walk formulations, and over an order of magnitude improvement over constrained matrix-triple-product formulations, in the context of real data sets.

Universal programmable quantum circuit schemes to emulate an operator

Unlike fixed designs, programmable circuit designs support an infinite number of operators. The functionality of a programmable circuit can be altered by simply changing the angle values of the rotation gates in the circuit. Here, we present a new quantum circuit design technique resulting in two general programmable circuit schemes. The circuit schemes can be used to simulate any given operator by setting the angle values in the circuit. This provides a fixed circuit design whose angles are determined from the elements of the given matrix-which can be non-unitary-in an efficient way. We also give both the classical and quantum complexity analysis for these circuits and show that the circuits require a few classical computations. For the electronic structure simulation on a quantum computer, one has to perform the following steps: prepare the initial wave function of the system; present the evolution operator U = e(-iHt) for a given atomic and molecular Hamiltonian H in terms of quantum gates array and apply the phase estimation algorithm to find the energy eigenvalues. Thus, in the circuit model of quantum computing for quantum chemistry, a crucial step is presenting the evolution operator for the atomic and molecular Hamiltonians in terms of quantum gate arrays. Since the presented circuit designs are independent from the matrix decomposition techniques and the global optimization processes used to find quantum circuits for a given operator, high accuracy simulations can be done for the unitary propagators of molecular Hamiltonians on quantum computers. As an example, we show how to build the circuit design for the hydrogen molecule.

Surfing the Network for Ranking by Multidamping

PageRank is one of the most commonly used techniques for ranking nodes in a network. It is a special case of a family of link-based rankings, commonly referred to as functional rankings. Functional rankings are computed as power series of a stochastic matrix derived from the adjacency matrix of the graph. This general formulation of functional rankings enables their use in diverse applications, ranging from traditional search applications to identification of spam and outliers in networks. This paper presents a novel algorithmic (re)formulation of commonly used functional rankings, such as LinearRank, TotalRank and Generalized Hyperbolic Rank. These rankings can be approximated by finite series representations. We prove that polynomials of stochastic matrices can be expressed as products of Google matrices (matrices having the form used in Googles original PageRank formulation). Individual matrices in these products are parameterized by different damping factors. For this reason, we refer to our formulation as multidamping. We demonstrate that multidamping has a number of desirable characteristics: (i) for problems such as finding the highest ranked pages, multidamping admits extremely fast approximate solutions; (ii) multidamping provides an intuitive interpretation of existing functional rankings in terms of the surfing habits of model web users; (iii) multidamping provides a natural framework based on Monte Carlo type methods that have efficient parallel and distributed implementations. It also provides the basis for constructing new link-based rankings based on inhomogeneous products of Google matrices. We present algorithms for computing damping factors for existing functional rankings analytically and numerically. We validate various benefits of multidamping on a number of real datasets.

Concurrent programming constructs for parallel MPI applications

Concurrency and parallelism have long been viewed as important, but somewhat distinct concepts. While concurrency is extensively used to amortize latency (for example, in web- and database-servers, user interfaces, etc.), parallelism is traditionally used to enhance performance through execution on multiple functional units. Motivated by an evolving application mix and trends in hardware architecture, there has been a push toward integrating traditional programming models for concurrency and parallelism. Use of conventional threads APIs (POSIX, OpenMP) with messaging libraries (MPI), however, leads to significant programmability concerns, owing primarily to their disparate programming models. In this paper, we describe a novel API and associated runtime for concurrent programming, called MPI Threads (MPIT), which provides a portable and reliable abstraction of low-level threading facilities. We describe various design decisions in MPIT, their underlying motivation, and associated semantics. We provide performance measurements for our prototype implementation to quantify overheads associated with various operations. Finally, we discuss two real-world use cases: an asynchronous message queue and a parallel information retrieval system. We demonstrate that MPIT provides a versatile, low overhead programming model that can be leveraged to program large parallel ensembles.

Jylab: A System for Portable Scientific Computing over Distributed Platforms

Jylab is a portable and flexible scientific computing system favoring extensibility. It provides users with a scripting language and a core set of libraries implementing numerical linear algebra routines, communication models, interactive visualization and computer algebra system capabilities. It thus enables the development of scientific applications involving numerics, graphics and symbolics over distributed computing platforms. It is enhanced with a collection of extension packages that support Grid computing and implement Web search engine and Web graph analysis frameworks. Building a multithreaded Internet algorithmics application in Jylab is fully presented.

Context-Specific Recommendation System for Predicting Similar PubMed Articles

Prioritizing a database of items in response to a given query object is a fundamental task in information retrieval and machine learning. We examine a specific realization of this problem in the context of a collection of biomedical articles. Given a query PubMed article, we investigate the problem of identifying and ranking recommended papers that are topically related to the query article. The two major classes of existing methods for this task are based on Natural Language Processing (NLP) techniques (including algebraic analyses), and those that incorporate structural information among articles, such as their co-citation networks or content similarity. In this paper, we propose a statistically rigorous method, called Context Specific Recommendation System (CSRS), along with associated algorithmic machinery to integrate structural and context-based sources of information to construct a single context-specific interaction network. We utilize this specialized network to rank papers (nodes) in terms of their similarity to query papers. Using a manually curated dataset of PubMed articles, we show that our method significantly outperforms other methods based on either the citation networks or content similarity of articles. Our methods provide a general framework that can be used to integrate other types of relationships into the recommendation process.

Grid-Enabling a Problem Solving Environment: Implementation and Everyday Use

We describe a simple, yet powerful API for accessing and using Grid resources from within Jylab, a novel, extensible scientific computing workbench consisting of a suite of open-source Java libraries scriptable through a Jython interpreter. The API provides a Java-based, Python-scriptable interactive environment and aims to simplify Grid application development and use. We demonstrate the utilization of the API in the context of an application from Internet algorithmics, specifically creating an index of crawled Web pages and using it for link-based ranking calculations (PageRank) and search queries.